-
Notifications
You must be signed in to change notification settings - Fork 11
/
Copy pathanalysis_code.R
1619 lines (1360 loc) · 67.8 KB
/
analysis_code.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# CODE FOR DERFINDER PAPER
# uses beta version of derfinder package
#########################################
#########################################
##### chunk 1: male vs. female analysis #
#########################################
#########################################
print(Sys.time())
library(devtools)
install_github('derfinder', 'alyssafrazee') # beta version
library(derfinder)
dbfile = "Y-tophat-revised.db"
tablename = "chrY"
textfile = "tophatY-updated"
# create database:
makeDb(dbfile = dbfile, tablename = tablename, textfile = textfile, cutoff = 5)
sex = c(1,1,0,0,1,1,0,0,1,1,1,1,0,0,1) #based on the samples in textfile
limma.input = getLimmaInput(dbfile = dbfile, tablename = tablename, group = sex, nonzero = TRUE)
pos = limma.input$pos
save(pos, file="posY-rev.rda")
# get the moderated t stats and fold changes:
tstats = getTstats(fit = limma.input$ebobject, trend = TRUE)
tt = tstats$tt
logfchange = tstats$logfchange
save(tt,file="ttY-rev.rda")
# fit the HMM:
find.them = getParams(tt)
regions = getRegions(method = "HMM", chromosome = "Y", pos = pos, tstats = tt, stateprobs = find.them$stateprobs, params = find.them$params, includet = TRUE, includefchange = TRUE, fchange = logfchange)
# merge the regions:
regions.merged.y = mergeRegions(regions$states)
save("regions.merged.y",file="Ychrom-regions-merged-new-rev.rda")
# get the p-values:
##### define p-value function to print status messages & the total number of null statistics, which we needed for one of the reviewer responses.
get.pvals <- function (regions, dbfile, tablename, num.perms = 1, group, est.params, chromosome, colsubset = c(-1)){
nullstats = NULL
for (i in 1:num.perms) {
group.permute = sample(group)
print(paste("starting iteration",i))
limma.input = getLimmaInput(dbfile = dbfile, tablename = tablename,group = group.permute, colsubset = colsubset, nonzero = TRUE)
tstats = getTstats(fit = limma.input$ebobject, trend = TRUE)
tt = tstats$tt
logfchange = tstats$logfchange
regions.null = getRegions(method = "HMM", chromosome = chromosome,
pos = limma.input$pos, tstats = tt, stateprobs = est.params$stateprobs,
params = est.params$params, includet = TRUE, includefchange = TRUE,
fchange = logfchange)
nullstats = append(nullstats, regions.null$states$mean.t[regions.null$states$state == 3 | regions.null$states$state == 4])
}
print("Number of null stats:")
print(length(nullstats))
pvals = rep(NA, dim(regions)[1])
for (k in which(regions$state == 3 | regions$state == 4)) {
pvals[k] = (sum(abs(nullstats) > abs(regions$mean.t[k]))+1)/(length(nullstats)+1)
}
return(pvals)
}
pvals = get.pvals(regions = regions.merged.y, dbfile = dbfile, tablename = tablename, num.perms = 10, group = sex, est.params = find.them, chromosome = "Y")
save(pvals, file="Y-pvals-new-rev.rda")
# get the flags:
exons = getAnnotation("hg19","knownGene")
myflags = getFlags(regions = regions.merged.y, exons, "chrY", pctcut = 0.8)
save(myflags, file="Y-flags-new-rev.rda")
#########################################
#########################################
##### chunk 2: male vs. male analysis ###
#########################################
#########################################
dbfile = "Y-tophat-revised.db"
tablename = "chrY"
textfile = "tophatY-updated"
# create database: (ALREADY MADE IN CHUNK 1)
#makeDb(dbfile = dbfile, tablename = tablename, textfile = textfile, cutoff = 5)
to.use = c(2,3,6,7,10,11,12,13,16)
set.seed(651)
sex = sample(c(rep(1,5),rep(0,4))) #randomly assign groups of males
limma.input = getLimmaInput(dbfile = dbfile, tablename = tablename, group = sex, colsubset = to.use, nonzero = TRUE)
pos = limma.input$pos
# get the moderated t stats and fold changes:
tstats = getTstats(fit = limma.input$ebobject, trend = TRUE)
tt = tstats$tt
logfchange = tstats$logfchange
save(tt,file="ttY-men-rev.rda")
# fit the HMM:
find.them = getParams(tt)
regions = getRegions(method = "HMM", chromosome = "Y", pos = pos, tstats = tt, stateprobs = find.them$stateprobs, params = find.them$params, includet = TRUE, includefchange = TRUE, fchange = logfchange)
# merge the regions:
regions.merged.y = mergeRegions(regions$states)
save("regions.merged.y",file="Ychrom-regions-merged-men-rev.rda")
# get the p-values: (uses same "debugging get.pvals" as above)
pvals = get.pvals(regions = regions.merged.y, dbfile =dbfile,tablename = tablename, num.perms = 10, group = sex, est.params = find.them, chromosome = "Y", colsubset = to.use)
save(pvals, file="Y-pvals-men-rev.rda")
#########################################
#########################################
##### chunk 3: analyze results ##########
#########################################
#########################################
## load in the exon annotation (ENSEMBL GRCh37)
library(biomaRt)
ensembl = useMart("ensembl")
listDatasets(ensembl)[which(listDatasets(ensembl)$dataset == "hsapiens_gene_ensembl"),]
### VERSION = GRCh37.p12
ensembl = useDataset("hsapiens_gene_ensembl", mart = ensembl)
filters = listFilters(ensembl)
attributes = listAttributes(ensembl)
ensexons = getBM(attributes=c("ensembl_gene_id", "ensembl_transcript_id",
"ensembl_exon_id", "chromosome_name", "exon_chrom_start",
"exon_chrom_end", "strand"), mart=ensembl)
# ^takes a few minutes
names(ensexons)[3] = "exon_id"
names(ensexons)[1] = "geneName"
names(ensexons)[4] = "chr"
names(ensexons)[5] = "start"
names(ensexons)[6] = "end"
ensYexons = subset(ensexons, chr=="Y")
nrow(ensYexons) #3748 exons, some duplicated.
ensYexons = ensYexons[!duplicated(ensYexons$exon_id),]
save(ensYexons, file="ensYexons.rda")
#################################################
##### chunk 3a: load resuls from DER Finder #####
#################################################
## load results from our pipeline:
load("ensYexons.rda")
load("Ychrom-regions-merged-men-rev.rda") #regions.merged.y
regions.merged.men = regions.merged.y
load("Ychrom-regions-merged-new-rev.rda") #also regions.merged.y
load("Y-pvals-men-rev.rda") #pvals
pvals.men = pvals
load("Y-pvals-new-rev.rda") #also pvals
regions = data.frame(regions.merged.y,pvals,qvals = p.adjust(pvals, method="fdr"))
regions.men = data.frame(regions.merged.men,pvals = pvals.men, qvals = p.adjust(pvals.men, method="fdr"))
### P VALUE HISTOGRAMS - check
hist(pvals,breaks=30,col="gray70",main="Y chromosome p values: our method",xlab="p values") #yes.
hist(pvals.men,breaks=30,col="gray70",main="Y chromosome p values: our method, men only",xlab="p values")
### get data frame of only DE regions (with p and q-values)
ders = subset(regions,state==3|state==4)
ders.men = subset(regions.men, state==3|state==4)
# re-create myflags with ensembl exons
myflags = getFlags(regions = regions.merged.y, ensYexons, "Y", pctcut = 0.8)
myflags.men = getFlags(regions = regions.merged.men, ensYexons, "Y", pctcut = 0.8)
save(myflags, file="Y-flags-ensembl.rda")
save(myflags.men, file="Y-flags-men-ensembl.rda")
# if flags already created:
# load("Y-flags-ensembl.rda")
# load("Y-flags-men-ensembl.rda")
ders = data.frame(ders,flag = myflags$flags)
ders.men = data.frame(ders.men,flag=myflags.men$flags)
# line up exons with p-values (i.e., get rid of the list situation)
info.lengths = sapply(myflags$flag.info, length, USE.NAMES=FALSE)
qvals.ex = ex.names = ex.pct = ex.class = list()
for(i in 1:length(myflags$flags)){
qvals.ex[[i]] = rep(ders$qvals[i],info.lengths[i])
ex.names[[i]] = myflags$flag.info[[i]]
ex.pct[[i]] = myflags$percent.exon[[i]]
ex.class[[i]] = rep(myflags$flags[i],info.lengths[i])
}
qvals.ex = unlist(qvals.ex)
ex.names = unlist(ex.names)
ex.class = unlist(ex.class)
ex.pct = unlist(ex.pct)
myinfo = data.frame(qvals.ex, ex.names, ex.class, ex.pct, stringsAsFactors=FALSE)
write.table(myinfo, file="myinfo_ensembl.txt",row.names=F,quote=F,sep="\t")
# fix places where exons are covered by >1 different region
#### RULES: if different parts of the same exon were overlapped
#### by different DE regions, that exon was reduced to just one
#### row, and the region was then flagged as DE if those combined
#### regions overlapped the exon by 80%. The q-value of
#### the largest region making up that exon's covering was taken
#### as the q-value for that exon.
#### (So if an exon was at least 80% overlapped by DERs
#### with q<0.05, it was called DE at the 0.05 level.)
rep_exons = names(table(myinfo$ex.names))[table(myinfo$ex.names)>1]
repsub = subset(myinfo, ex.names %in% rep_exons)
percentsplit = split(repsub$ex.pct, repsub$ex.names)
qsplit = split(repsub$qvals.ex, repsub$ex.names)
sum(names(qsplit) != rep_exons) ## 0; GOOD
newpct = sapply(percentsplit, sum, USE.NAMES=FALSE)
newq = sapply(1:length(qsplit), function(i){
qsplit[[i]][which.max(percentsplit[[i]])]
}, USE.NAMES=FALSE)
newclass = ifelse(newpct>=0.8, "DE exons(s)", NA)
myinfo.updated = subset(myinfo, !(ex.names %in% rep_exons))
myinfo.updated = rbind(myinfo.updated, data.frame(qvals.ex = newq,
ex.names=names(qsplit), ex.class=newclass, ex.pct=newpct))
rownames(myinfo.updated) = NULL
myinfo.sub = subset(myinfo.updated, ex.class!="novel")
nrow(myinfo.sub) #467 exons
checkunique = function(x){
sum(ensYexons$exon_id==x)
}
checkexons = unlist(lapply(myinfo.updated$ex.names, checkunique))
table(checkexons) #no id issues
foo = which(unlist(lapply(myinfo.updated$ex.names, checkunique))>1)
# same thing for men:
info.lengths.m = sapply(myflags.men$flag.info, length, USE.NAMES=FALSE)
qvals.ex.m = ex.names.m = ex.pct.m = ex.class.m = list()
for(i in 1:length(myflags.men$flags)){
qvals.ex.m[[i]] = rep(ders.men$qvals[i],info.lengths.m[i])
ex.names.m[[i]] = myflags.men$flag.info[[i]]
ex.pct.m[[i]] = myflags.men$percent.exon[[i]]
ex.class.m[[i]] = rep(myflags.men$flags[i],info.lengths.m[i])
}
qvals.ex.m = unlist(qvals.ex.m)
ex.names.m = unlist(ex.names.m)
ex.class.m = unlist(ex.class.m)
ex.pct.m = unlist(ex.pct.m)
myinfo.m = data.frame(qvals.ex.m, ex.names.m, ex.class.m, ex.pct.m)
min(myinfo.m$qvals.ex.m) #the min q-value here is 0.86, so nothing will show up in the table.
#################################################
##### chunk 3a: load resuls from Cufflinks ######
#################################################
# load 'isoform.diff' files from Cuffdiff
load("tx.sex.cuff-UPDATED.rda")
load("tx.null.cuff-UPDATED.rda")
tx.sex.cuff.y.ok = subset(tx.sex.cuff,substr(locus,1,1)=="Y" & status=="OK")
## recalibrate q values (only using Y chromosome)
tx.sex.cuff.y.ok$qvalue = p.adjust(tx.sex.cuff.y.ok$pvalue,method="fdr")
cuff.t = tx.sex.cuff.y.ok$test_stat
# fix some very, very large (10e308) test stats...
cuff.t = ifelse(cuff.t>20,20,cuff.t)
cuff.t = ifelse(cuff.t<(-20),-20,cuff.t)
cuff.fchange = tx.sex.cuff.y.ok$log2fchange
cuff.fchange = ifelse(cuff.fchange>20,20,cuff.fchange)
cuff.fchange = ifelse(cuff.fchange<(-20),-20,cuff.fchange)
cuff.t[which(cuff.fchange==-20)] = 20
tx.null.cuff.y.ok = subset(tx.null.cuff,substr(locus,1,1)=="Y" & status=="OK")
# recalibrate q:
tx.null.cuff.y.ok$qvalue = p.adjust(tx.null.cuff.y.ok$pvalue,method="fdr")
tx.sex.cuff.y = subset(tx.sex.cuff,substr(locus,1,1)=="Y")
tx.null.cuff.y = subset(tx.null.cuff,substr(locus,1,1)=="Y")
### P VALUE HISTOGRAMS ### - just to check
hist(tx.sex.cuff.y.ok$pvalue, col="gray70",breaks=30, xlab="p values",main="Y chromosome p values: Cufflinks")
hist(tx.null.cuff.y.ok$pvalue, col="gray70",breaks=30, xlab="p values",main="Y chromosome null p values: Cufflinks")
# merged.gtf results, containing the exons of each transcript:
# (result of Cuffmerge)
# (this is for the supplemental table)
merged.gtf = read.table("normal-comparisons/merged.gtf",sep="\t")
merged.gtf$V9 = as.character(merged.gtf$V9)
summary(merged.gtf$V3) #all exon- good.
elts = unlist(strsplit(merged.gtf$V9,split="; "))
isstandard = function(string){
isgid = substring(string,1,7)=="gene_id"
istx = substring(string,1,13)=="transcript_id"
isexnum = substring(string,1,11)=="exon_number"
isoid = substring(string,1,3)=="oId"
istssid = substring(string,1,6)=="tss_id"
return(isgid|istx|isexnum|isoid|istssid)
}
elts = elts[-which(!isstandard(elts))] #this should work...
merged.gtf.new = merged.gtf[,-9]
names(merged.gtf.new) = c("chr","software","feature","start","end","score","strand","frame")
striplabel = function(x){
splitme = strsplit(x,split=" ")[[1]][2]
return(splitme)
}
elts = unlist(lapply(elts,striplabel))
eltsdf = matrix(elts,ncol=5,byrow=TRUE)
merged.gtf.new = data.frame(chr=merged.gtf.new$chr,start=merged.gtf.new$start, end=merged.gtf.new$end, feature=merged.gtf.new$feature, geneid=eltsdf[,1], txid=eltsdf[,2], exnum = as.numeric(eltsdf[,3]))
merged.gtf.y = subset(merged.gtf.new, chr=="Y")
merged.gtf.y = merged.gtf.y[order(merged.gtf.y$start),] # we want to use THIS - just add in qvals & test status
qvalue = status = NULL
for(i in 1:dim(merged.gtf.y)[1]){
print(i)
ind = which(tx.sex.cuff.y$tx_id==merged.gtf.y$txid[i])
status[i] = as.character(tx.sex.cuff.y$status[ind])
if(status[i]=="OK"){
ind2 = which(tx.sex.cuff.y.ok$tx_id==merged.gtf.y$txid[i])
qvalue[i] = tx.sex.cuff.y.ok$qvalue[ind2]
}
if(status[i]!="OK") qvalue[i] = 1
} #takes a while (~ 1 hour)
merged.gtf.y = data.frame(merged.gtf.y, qvalue, status) #sweet.
save(merged.gtf.y,file="cuffdiff-sex/merged.gtf.y.rda")
### IF ABOVE ALREADY DONE:
load("cuffdiff-sex/merged.gtf.y.rda")
merged.gtf.y.ok = subset(merged.gtf.y,status=="OK")
unique.transcripts = merged.gtf.y.ok[!duplicated(as.character(merged.gtf.y.ok$txid)),]
dim(unique.transcripts) #808 rows
### now get the merged file for the men only: (repeat previous process)
merged.gtf.men = read.table("merged-men.gtf",sep="\t")
merged.gtf.men$V9 = as.character(merged.gtf.men$V9)
summary(merged.gtf.men$V3) #all exon- good.
elts.men = unlist(strsplit(merged.gtf.men$V9,split="; "))
elts.men = elts.men[-which(!isstandard(elts.men))] #this should work...
merged.gtf.men.new = merged.gtf.men[,-9]
names(merged.gtf.men.new) = c("chr","software","feature","start","end","score","strand","frame")
elts.men = unlist(lapply(elts.men,striplabel))
eltsdf.men = matrix(elts.men,ncol=5,byrow=TRUE)
merged.gtf.men.new = data.frame(chr=merged.gtf.men.new$chr,start=merged.gtf.men.new$start, end=merged.gtf.men.new$end, feature=merged.gtf.men.new$feature, geneid=eltsdf.men[,1], txid=eltsdf.men[,2], exnum = as.numeric(eltsdf.men[,3]))
merged.gtf.men.y = subset(merged.gtf.men.new, chr=="Y")
merged.gtf.men.y = merged.gtf.men.y[order(merged.gtf.men.y$start),]
qvalue.m = status.m = NULL
dim(merged.gtf.men.y) #1961 rows
for(i in 1:dim(merged.gtf.men.y)[1]){
print(i)
ind = which(tx.null.cuff.y$tx_id==merged.gtf.men.y$txid[i])
status.m[i] = as.character(tx.null.cuff.y$status[ind])
if(status.m[i]=="OK"){
ind2 = which(tx.null.cuff.y.ok$tx_id==merged.gtf.men.y$txid[i])
qvalue.m[i] = tx.null.cuff.y.ok$qvalue[ind2]
}
if(status.m[i]!="OK") qvalue.m[i] = 1
} #again, takes a while (maybe an hour)
merged.gtf.men.y = data.frame(merged.gtf.men.y, qvalue=qvalue.m, status=status.m) #sweet.
save(merged.gtf.men.y,file="cuffdiff-males/merged.gtf.men.y.rda")
##### IF ABOVE ALREADY DONE:
load("cuffdiff-males/merged.gtf.men.y.rda")
merged.gtf.men.y.ok = subset(merged.gtf.men.y,status=="OK")
unique.transcripts.men = merged.gtf.men.y.ok[!duplicated(as.character(merged.gtf.men.y.ok$txid)),]
dim(unique.transcripts.men) #818 rows
## how many of the cufflinks (or other) DE transcripts are 80% covered by DERs from our pipeline?
doweagree = function(qcut, unique.transcripts, ders, txinfo, pctcut = 0.8){
#unique.transcripts is a data frame containing txid and q-value (one per transcript)
#txinfo is the merged.gtf file, containing the exon locations and txids. can have multiple rows per transcript, but only one row per exon
k=0
percent.covered = NULL
dercands = subset(ders,qvals<=qcut)
for(i in which(unique.transcripts$qvalue<qcut)){
print("hey")
print(i)
k=k+1
tname = as.character(unique.transcripts$txid[i])
theexons = subset(txinfo, txid==tname)
firstderind = findInterval(theexons$start[1], dercands$start)
lastderind = findInterval(theexons$end[dim(theexons)[1]], dercands$start)
if(firstderind==0) firstderind = 1
if(lastderind==0){percent.covered[k] = 0; next}
these.ders = dercands[firstderind:lastderind,]
transcript.pos = NULL
for(j in 1:dim(theexons)[1]){
transcript.pos = append(transcript.pos, c(theexons$start[j]:theexons$end[j]))
}
ders.pos = NULL
for(j in 1:dim(these.ders)[1]){
ders.pos = append(ders.pos, c(these.ders$start[j]:these.ders$end[j]))
}
percent.covered[k] = sum(transcript.pos %in% ders.pos)/length(transcript.pos)
}
numcovered = sum(percent.covered>pctcut)
return(numcovered)
}
## Among our regions, how many are 80% covered by a DE cufflinks transcript?
dotheyagree = function(qcut, ychr.cuff, ders){
#ychr.cuff is the merged.gtf file, with q-values and all exons, from cufflinks
#ders is our ders data frame.
require(IRanges)
cover80pct = NULL
k = 0
ychr.cuff.sig = subset(ychr.cuff,qvalue<qcut)
for(i in which(ders$qvals<qcut)){
print(i)
k = k+1
cuffind = findInterval(ders$start[i], ychr.cuff.sig$start)
endind = findInterval(ders$end[i], ychr.cuff.sig$start)
if(endind==0){
cover80pct[k] = 0
next
}
endInendind = ders$end[i] %in% c((ychr.cuff.sig$start[endind]):(ychr.cuff.sig$end[endind]))
if(cuffind==0){
x = IRanges(start = ychr.cuff$start[1:endind], end = ychr.cuff$end[1:endind])
x = reduce(x)
tosum = width(x)
if(endInendind){tosum[length(tosum)] = ders$end[i] - start(x)[length(tosum)] + 1}
numcovered = sum(tosum)
percent.covered = numcovered/(ders$length[i])
cover80pct[k] = ifelse(percent.covered>=0.8,1,0)
next
} #end cuffind==0 case
startIncuffind = ders$start[i] %in% c((ychr.cuff.sig$start[cuffind]):(ychr.cuff.sig$end[cuffind]))
if(cuffind==endind){
if(startIncuffind & endInendind){
cover80pct[k] = 1
next
}
if(startIncuffind & !endInendind){
numcovered = ychr.cuff$end[cuffind] - ders$start[i] + 1
percent.covered = numcovered/(ders$length[i])
cover80pct[k] = ifelse(percent.covered>=0.8,1,0)
next
}
if(!startIncuffind){
cover80pct[k] = 0
next
}
} #end equal case
if(cuffind!=endind){
x = IRanges(start = ychr.cuff$start[cuffind:endind], end = ychr.cuff$end[cuffind:endind])
x = reduce(x)
tosum = width(x)
if(endInendind){tosum[length(tosum)] = ders$end[i] - start(x)[length(tosum)] + 1}
if(startIncuffind){tosum[1] = end(x)[1] - ders$start[i] + 1}
if(!startIncuffind){tosum[1] = 0}
numcovered = sum(tosum)
percent.covered = numcovered/(ders$length[i])
cover80pct[k] = ifelse(percent.covered>=0.8,1,0)
} #end unequal case
} #end for i
return(sum(cover80pct))
} #end function
#### CREATE TABLE comparing our results to cufflinks results
cufftable = matrix(NA, ncol=9, nrow=17)
cufftable = as.data.frame(cufftable)
names(cufftable) = c("q","numregions","numtranscripts","doweagree","dotheyagree","numregions.men","numtranscripts.men","doweagree.men","dotheyagree.men")
cufftable$q = seq(0,0.8,by=0.05)
for(r in 1:17){
cufftable$numregions[r] = length(which(ders$qvals<cufftable$q[r]))
cufftable$numtranscripts[r] = length(which(unique.transcripts$qvalue<cufftable$q[r]))
if(cufftable$numregions[r]==0 & cufftable$numtranscripts[r]==0){
cufftable$dotheyagree[r] = NA
cufftable$doweagree[r] = NA
}else if(cufftable$numtranscripts[r]==0){
cufftable$dotheyagree[r] = 0
cufftable$doweagree[r] = NA
}else if(cufftable$numregions[r]==0){
cufftable$doweagree[r] = 0
cufftable$dotheyagree[r] = NA
}else{
cufftable$dotheyagree[r] = dotheyagree(cufftable$q[r], ychr.cuff = merged.gtf.y.ok, ders = ders)
cufftable$doweagree[r] = doweagree(cufftable$q[r], unique.transcripts = unique.transcripts, txinfo = merged.gtf.y.ok, ders = ders)
}
cufftable$numregions.men[r] = length(which(ders.men$qvals<cufftable$q[r]))
cufftable$numtranscripts.men[r] = length(which(unique.transcripts.men$qvalue<cufftable$q[r]))
if(cufftable$numregions.men[r]==0 & cufftable$numtranscripts.men[r]==0){
cufftable$dotheyagree.men[r] = NA
cufftable$doweagree.men[r] = NA
}else if(cufftable$numtranscripts.men[r]==0){
cufftable$dotheyagree.men[r] = 0
cufftable$doweagree.men[r] = NA
}else if(cufftable$numregions.men[r]==0){
cufftable$doweagree.men[r] = 0
cufftable$dotheyagree.men[r] = NA
}else{
cufftable$dotheyagree.men[r] = dotheyagree(cufftable$q[r], ychr.cuff = merged.gtf.y.men.ok, ders = ders.men)
cufftable$doweagree[r] = doweagree(cufftable$q[r], unique.transcripts = unique.transcripts.men, txinf = merged.gtf.men.y.ok, ders = ders.men)
}
}
cufftable ### supplementary table 2
save(cufftable, file="cufftable-correct-rev.rda")
## if above already done:
load("cufftable-correct-rev.rda")
cufftable
r=11 #q=0.5
doweagree(cufftable$q[r], unique.transcripts = unique.transcripts, txinfo = merged.gtf.y.ok, ders = ders, pctcut = 0.25) #99
sum(unique.transcripts$qvalue<=0.5) #758
99/758 #13.1 %
doweagree(cufftable$q[r], unique.transcripts = unique.transcripts, txinfo = merged.gtf.y.ok, ders = ders, pctcut = 0) #135
135/758 #17.8%
#####################################################
##### chunk 3b: get resuls from DESeq & EdgeR #######
#####################################################
###############################################
###### chunk 3b.1 -- make the exon count tables
###############################################
library(Rsamtools)
library(GenomicFeatures)
load("ensYexons.rda")
# take out same exon annotated in different transcripts:
yx = unique(ensYexons[,-2])
exons = GRanges(seqnames = Rle("Y"),
ranges = IRanges(start = yx$start, end = yx$end),
strand = Rle(yx$strand),
exon = yx$exon_id,
gene = yx$geneName)
strand(exons) = "*"
countmat = NULL
bamFls = rep(NA,15)
sampleNums = c(1,11,23,2,32,33,3,40,42,43,47,53,55,56,58)
for(i in 1:15){
bamFls[i] = paste("orbFrontalF",sampleNums[i],"/tophat/accepted_hits.bam",sep="")
aln = readGAlignments(bamFls[i])
strand(aln) = "*"
counts = summarizeOverlaps(features = exons, reads = aln, mode = "Union")
countmat = cbind(countmat, assays(counts)$count)
}
rownames(countmat) = ensYexons$exon_id
colnames(countmat) = paste0("orbFrontalF",sampleNums)
save(countmat, file="ensembl-exoncounts.rda")
###############################################
###### chunk 3b.2 -- run DESeq and EdgeR ######
###############################################
library(edgeR)
library(DESeq)
library(GenomicRanges)
# load per-exon counts:
load("ensembl-exoncounts.rda") #countmat
length(unique(rownames(countmat))) == nrow(countmat)
#TRUE, passes test.
# label samples:
sex = c(1,1,0,0,1,0,1,0,1,1,1,1,0,0,1)
# make an updated version of edgeR's function, so it will return the group abundances:
exactTest.more = function (object, pair = 1:2, dispersion = "auto", rejection.region = "doubletail", big.count = 900, prior.count.total = 0.5)
{
if (!is(object, "DGEList"))
stop("Currently only supports DGEList objects as the object argument.")
if (length(pair) != 2)
stop("Pair must be of length 2.")
rejection.region <- match.arg(rejection.region, c("doubletail", "deviance", "smallp"))
group <- as.factor(object$samples$group)
levs.group <- levels(group)
if (is.numeric(pair))
pair <- levs.group[pair]
else pair <- as.character(pair)
if (!all(pair %in% levs.group))
stop("At least one element of given pair is not a group.\n Groups are: ",
paste(levs.group, collapse = " "), "\n")
if (is.null(dispersion))
dispersion <- "auto"
if (is.character(dispersion)) {
dispersion <- match.arg(dispersion, c("auto", "common", "trended", "tagwise"))
dispersion <- switch(dispersion, common = object$common.dispersion, trended = object$trended.dispersion, tagwise = object$tagwise.dispersion, auto = getDispersion(object))
if (is.null(dispersion))
stop("specified dispersion not found in object")
}
ldisp <- length(dispersion)
ntags <- nrow(object$counts)
if (ldisp != 1 && ldisp != ntags)
stop("Dispersion provided by user must have length either 1 or the number of tags in the DGEList object.")
if (ldisp == 1)
dispersion <- rep(dispersion, ntags)
group <- as.character(group)
j <- group %in% pair
y <- object$counts[, j, drop = FALSE]
lib.size <- object$samples$lib.size[j]
norm.factors <- object$samples$norm.factors[j]
group <- group[j]
if (is.null(rownames(y)))
rownames(y) <- paste("tag", 1:ntags, sep = ".")
lib.size <- lib.size * norm.factors
offset <- log(lib.size)
lib.size.average <- exp(mean(offset))
abundance <- mglmOneGroup(y, dispersion = dispersion, offset = offset)
logCPM <- (abundance + log(1e+06))/log(2)
prior.count <- lib.size
prior.count <- prior.count.total * prior.count/sum(prior.count)
j1 <- group == pair[1]
n1 <- sum(j1)
if (n1 == 0)
stop("No libraries for", pair[1])
y1 <- y[, j1, drop = FALSE]
abundance1 <- mglmOneGroup(y1 + matrix(prior.count[j1], ntags, n1, byrow = TRUE), offset = offset[j1])
j2 <- group == pair[2]
n2 <- sum(j2)
if (n1 == 0)
stop("No libraries for", pair[2])
y2 <- y[, j2, drop = FALSE]
abundance2 <- mglmOneGroup(y2 + matrix(prior.count[j2], ntags, n2, byrow = TRUE), offset = offset[j2])
logFC <- (abundance2 - abundance1)/log(2)
e <- exp(abundance)
input.mean <- matrix(e, ntags, n1)
output.mean <- input.mean * lib.size.average
input.mean <- t(t(input.mean) * lib.size[j1])
y1 <- q2qnbinom(y1, input.mean = input.mean, output.mean = output.mean, dispersion = dispersion)
input.mean <- matrix(e, ntags, n2)
output.mean <- input.mean * lib.size.average
input.mean <- t(t(input.mean) * lib.size[j2])
y2 <- q2qnbinom(y2, input.mean = input.mean, output.mean = output.mean, dispersion = dispersion)
exact.pvals <- switch(rejection.region, doubletail = exactTestDoubleTail(y1, y2, dispersion = dispersion, big.count = big.count), deviance = exactTestByDeviance(y1, y2, dispersion = dispersion, big.count = big.count), smallp = exactTestBySmallP(y1, y2, dispersion = dispersion, big.count = big.count))
de.out <- data.frame(logFC = logFC, logCPM = logCPM, PValue = exact.pvals, ab1 = exp(abundance1), ab2=exp(abundance2))
rownames(de.out) <- rownames(object$counts)
new("DGEExact", list(table = de.out, comparison = pair, genes = object$genes))
}
### edgeR - men v women
y = DGEList(counts = countmat, group = sex)
y = estimateCommonDisp(y)
y = estimateTagwiseDisp(y)
et = exactTest.more(y)
edger.table = topTags(et, n = dim(et$table)[1], adjust.method = "BH")
head(edger.table$table,n=20)
edger.results = edger.table$table[is.finite(edger.table$table$logCPM),] ## remove exons with 0 counts in both groups
pdf(file="phist-edgeR-sex.pdf")
hist(edger.results$PValue, col="gray70",xlab="p value",main="EdgeR p values - male vs. female, Y", breaks=30)
dev.off()
save(edger.results,file="edger.results_ensembl.rda") ## in working directory
# edgeR - men v. men
men.table = countmat[,which(sex==1)]
set.seed(651)
y2 = DGEList(counts = men.table, group = sample(c(rep(1,5),rep(0,4))) )
y2 = estimateCommonDisp(y2)
y2 = estimateTagwiseDisp(y2)
et2 = exactTest.more(y2)
edger.table.men = topTags(et2, n = dim(et2$table)[1], adjust.method = "BH")
head(edger.table.men$table,n=10)
edger.results.men = edger.table.men$table[is.finite(edger.table$table$logCPM),]
pdf(file="phist-edgeR-men.pdf")
hist(edger.results.men$PValue, col="gray70",xlab="p value",main="Edger p values: men only, Y")
dev.off()
save(edger.results.men,file="edger.results.men_ensembl.rda")
# DESeq - men v women
cds = newCountDataSet(countmat, sex)
cds = estimateSizeFactors(cds)
sizeFactors(cds)
cds = estimateDispersions(cds)
des.res = nbinomTest(cds, "1","0")
head(des.res)
deseq.table = des.res[order(des.res$padj),]
pdf(file="phist-DESeq-sex.pdf")
hist(deseq.table$pval, col="gray70",xlab="p value",main="DESeq p values: male vs. female, Y", breaks=30)
dev.off()
save(deseq.table,file="deseq.table_ensembl.rda")
# DESeq - men v men
set.seed(651)
cds2 = newCountDataSet(men.table, sample(c(rep(1,5),rep(0,4))))
cds2 = estimateSizeFactors(cds2)
cds2 = estimateDispersions(cds2)
des.res.men = nbinomTest(cds2,"1","0")
head(des.res.men)
deseq.table.men = des.res.men[order(des.res.men$padj),]
pdf(file="phist-DESeq-men.pdf")
hist(deseq.table.men$pval,col="gray70",xlab="p value",main="DESeq p values: men only, Y")
dev.off()
save(deseq.table.men,file="deseq.table.men_ensembl.rda")
###############################################
###### chunk 3b.3 -- compare to DER Finder ####
###############################################
###### make edger/deseq table: (supplementary table 3)
annottable = matrix(NA,ncol=9, nrow=17)
annottable = as.data.frame(annottable)
names(annottable) = c("q","numregions","num.exons.us", "num.exons.edgeR","num.exons.DESeq", "edger.us.agree", "deseq.us.agree", "deseq.edger.agree", "all.agree")
annottable$q = seq(0,0.8,by=0.05)
int2 = function(z) Reduce('intersect',z) # need to take intersection of more than 2 sets -- this is how you do it.
for(r in 1:17){
annottable$numregions[r] = sum(ders$qvals<annottable$q[r])
annottable$num.exons.us[r] = length(unique(myinfo.sub$ex.names[myinfo.sub$ex.pct>=0.8 & myinfo.sub$qvals.ex<annottable$q[r]]))
ourexons = unique(myinfo.sub$ex.names[myinfo.sub$ex.pct>=0.8 & myinfo.sub$qvals.ex<annottable$q[r]])
annottable$num.exons.edgeR[r] = sum(edger.results$FDR<annottable$q[r])
edgerexons = rownames(edger.results)[which(edger.results$FDR<annottable$q[r])]
# need to fix exon names (included gene names in edgeR/deseq tables)
ulength = length(edgerexons)
edgerexons = unlist(lapply(edgerexons, function(x) strsplit(x,"-")[[1]][1]))
ulength2 = length(unique(edgerexons))
if(ulength!=ulength2) print(paste("OH NO EDGER",r))
annottable$num.exons.DESeq[r] = length(which((deseq.table$padj<annottable$q[r])))
deseqexons = deseq.table$id[which((deseq.table$padj<annottable$q[r]))]
ulength = length(deseqexons)
deseqexons = unlist(lapply(deseqexons, function(x) strsplit(x,"-")[[1]][1]))
ulength2 = length(unique(deseqexons))
if(ulength!=ulength2) print(paste("OH NO DESEQ",r))
annottable$edger.us.agree[r] = length(intersect(edgerexons,ourexons))
annottable$deseq.us.agree[r] = length(intersect(deseqexons,ourexons))
annottable$deseq.edger.agree[r] = length(int2(list(deseqexons, edgerexons)))
annottable$all.agree[r] = length(int2(list(edgerexons,ourexons,deseqexons)))
}
annottable
length(setdiff(myinfo.sub$ex.names, rownames(edger.results))) ##345
## investigate a few of the exons that EdgeR/DESeq find, but we do not.
r=2 #set q=0.05
ourexons = unique(myinfo.sub$ex.names[myinfo.sub$ex.pct>=0.8 & myinfo.sub$qvals.ex<annottable$q[r]])
edgerexons = rownames(edger.results)[which(edger.results$FDR<annottable$q[r])]
deseqexons = deseq.table$id[which((deseq.table$padj<annottable$q[r]))]
edgeronly = setdiff(edgerexons,ourexons)
length(edgeronly) #47 exons
deseqonly = setdiff(deseqexons,ourexons)
length(deseqonly) #39 exons
length(union(edgeronly, deseqonly)) ##54 discovered only by edger or deseq
## setup for plotting:
gro = list()
gro$states = regions.merged.y
states.norle.temp = inverse.rle(list(lengths=regions.merged.y$length, values=regions.merged.y$state))
load("posY-rev.rda") # should already be loaded, but just in case
load("ttY-rev.rda") # should already be loaded, but just in case
states.norle.temp2 = states.norle.temp[pos-pos[1]+1]
gro$states.norle=data.frame(pos=pos, states=states.norle.temp2)
rm(states.norle.temp, states.norle.temp2);gc();gc();gc()
group = c(1,1,0,0,1,1,0,0,1,1,1,1,0,0,1)
group.l = ifelse(group==1,"male","female")
ensYexons.forplot = ensYexons
names(ensYexons.forplot)[1] = "geneName"
names(ensYexons.forplot)[3] = "name"
# plot those exons found only by edgeR/DEseq:
pdf("only_edger_ens.pdf")
for(ex in edgeronly){
plotExon(gro, exonname=ex, tstats=tt, pos=pos, annotation=ensYexons.forplot, counts="Y-tophat-revised.db", tabname="chrY", chromosome="Y", group=group.l, scalefac=32,ylim=c(4.5,7.5))
}
dev.off()
pdf("only_deseq_ens.pdf")
for(ex in deseqonly){
plotExon(gro, exonname=ex, tstats=tt, pos=pos, annotation=ensYexons.forplot, counts="Y-tophat-revised.db", tabname="chrY", chromosome="Y", group=group.l, scalefac=32,ylim=c(4.5,7.5))
}
dev.off()
length(intersect(deseqonly, edgeronly)) #32
### look at these for page 14 of DER Finder manuscript
# some miscellaneous statistics (included in results/text of manuscript)
sum(ders$qvals<0.05) #534
sum(ders$qvals<0.05 & ders$state==4) #6
length(which(ders$flag=="novel" & ders$qvals<0.05)) #280
summary(ders$length[which(ders$flag=="novel" & ders$qvals<0.05)]) #range: 1-3814
de.exons = myinfo.sub$ex.names[which(myinfo.sub$qvals.ex<0.05 & myinfo.sub$ex.pct>=0.8)]
length(unique(de.exons)) #411, and "novel" is not included.
genes.represented = ensYexons$geneName[which(ensYexons$exon_id %in% unique(de.exons))]
length(unique(genes.represented)) #33
min(ders.men$qvals) #0.86
min(tx.sex.cuff.y.ok$qvalue) #0.45
min(tx.null.cuff.y.ok$qvalue) #0.63
sum(tx.sex.cuff.y.ok$value_female==0 & tx.sex.cuff.y.ok$value_male!=0) #736
#########################################
#########################################
##### chunk 4: figures ##################
#########################################
#########################################
#### FIGURE 1
load("chr22exons.rda")
xx = exondata22[exondata22$exon_chrom_start>20936000 & exondata22$exon_chrom_end<20946000 & exondata22$exon_chrom_start < 20941000,]
dim(xx) #108 rows
length(unique(xx$ensembl_exon_id)) #40 unique exons by id
length(unique(xx$ensembl_transcript_id)) #15 transcripts
setEPS(width=12, height=6)
postscript("exons.eps")
#colored_exons = NULL
par(mfrow=c(1,2))
### transcript structures
xax = seq(min(xx$exon_chrom_start), max(xx$exon_chrom_end), by=1)
plot(xax, rep(0,length(xax)), ylim=c(0,17), type="n", xlab="Genomic Position", yaxt = "n", ylab="")
title("(a) Annotated Transcripts: Ensembl 61, Chromosome 22")
for(tx in unique(as.character(xx$ensembl_transcript_id))){
txind = which(unique(xx$ensembl_transcript_id)==tx)
gtsub = xx[xx$ensembl_transcript_id==tx,]
gtsub = gtsub[order(gtsub$exon_chrom_start),]
for(exind in 1:nrow(gtsub)){
#mycolor = ifelse(as.character(gtsub$ensembl_exon_id[exind]) %in% colored_exons, "white","gray60")
mycolor = "gray60"
polygon(x=c(gtsub$exon_chrom_start[exind],
gtsub$exon_chrom_start[exind],
gtsub$exon_chrom_end[exind],
gtsub$exon_chrom_end[exind]),
y=c(txind-0.4,txind+0.4,txind+0.4,txind-0.4),
col=mycolor)
if(exind != nrow(gtsub)) lines(c(gtsub$exon_chrom_end[exind],gtsub$exon_chrom_start[exind+1]),c(txind, txind), lty=2, col="gray60")
#colored_exons = append(as.character(gtsub$ensembl_exon_id[exind]), colored_exons)
}
}
#legend("bottomright", pch=15, col="gray60", "unique exon")
### overlapping exons
olap_exons = subset(xx, exon_chrom_start<20941000 & exon_chrom_start>20940500 &
exon_chrom_end<20942000 & exon_chrom_end>20941500)
olap_exons = olap_exons[which(!duplicated(olap_exons[,c(5,6)])),]
xax = seq(olap_exons$exon_chrom_start[1]-100, max(olap_exons$exon_chrom_end)+100, by=1)
plot(xax, rep(0,length(xax)), ylim=c(0.5,4.5),
type="n", xlab="Genomic Position", yaxt = "n", ylab="",
xlim=c(20941800, 20941950))
for(i in seq_along(olap_exons[,1])){
polygon(x=c(olap_exons$exon_chrom_start[i],
olap_exons$exon_chrom_start[i],
olap_exons$exon_chrom_end[i],
olap_exons$exon_chrom_end[i]),
y=c(i-0.4, i+0.4, i+0.4, i-0.4),
col="gray60")
}
title("(b) Close-up of Exon Annotation Differences")
dev.off()
# percentile vs. %MF plot [FIGURE 4]
regions.menA = regions.men
regions.A = regions
ders.men = subset(regions.men, state==3|state==4)
ders = subset(regions, state==3|state==4)
regions = ders
regions.men = ders.men
regions$exp = rep("sex",dim(regions)[1])
regions.men$exp = rep("men",dim(regions.men)[1])
us.t = c(regions$mean.t,regions.men$mean.t)
us.exp = c(regions$exp, regions.men$exp)
us.state = c(regions$state, regions.men$state)
us.length = c(regions$length, regions.men$length)
usp = regions$pvals
usp[regions$mean.t<0] = -usp[regions$mean.t<0]
unp = regions.men$pvals
unp[regions.men$mean.t<0] = -unp[regions.men$mean.t<0] # pos t means overexpressed in men
us.p = c(usp,unp)
us.res = data.frame(t = us.t, exp = us.exp, p=us.p)
us.res = us.res[us.length>=102,]
us.res = us.res[order(1/us.res$p,decreasing=T),]
# handle ties (randomly sample labels):
p.rle = rle(us.res$p)
p.rle = data.frame(lengths=p.rle$lengths,values=p.rle$values)
for(k in which(p.rle$lengths>1)){
inds = which(us.res$p==p.rle$values[k])
us.res$exp[inds] = sample(us.res$exp[inds])
}
us.pct = NULL
for(i in 1:nrow(us.res)){
us.pct[i] = sum(us.res$exp[1:i]=="sex")/i
}
t.percentile = NULL
for(i in 1:length(us.res$t)){ t.percentile[i] = 1-i/length(us.res$t)}
cuff.t = tx.sex.cuff.y.ok$test_stat
cuff.t = ifelse(cuff.t>20,20,cuff.t)
cuff.t = ifelse(cuff.t<(-20),-20,cuff.t)
cuff.fchange = tx.sex.cuff.y.ok$log2fchange
cuff.fchange = ifelse(cuff.fchange>20,20,cuff.fchange)
cuff.fchange = ifelse(cuff.fchange<(-20),-20,cuff.fchange)
cuff.t.m = tx.null.cuff.y.ok$test_stat
cuff.t.m = ifelse(cuff.t.m>20,20,cuff.t.m)
cuff.t.m = ifelse(cuff.t.m<(-20),-20,cuff.t.m)
cuff.fchange.m = tx.null.cuff.y.ok$log2fchange
cuff.fchange.m = ifelse(cuff.fchange.m>20,20,cuff.fchange.m)
cuff.fchange.m = ifelse(cuff.fchange.m<(-20),-20,cuff.fchange.m)
cuff.t.all = c(cuff.t,cuff.t.m)
cuff.fchange[which(tx.sex.cuff.y.ok$log2fchange<(-20))] = -20
cuff.fc.all = c(cuff.fchange, cuff.fchange.m)
cuff.p.tmp = tx.sex.cuff.y.ok$pvalue
# positive p-values
cuff.p.tmp[which(tx.sex.cuff.y.ok$test_stat<(-20) | (tx.sex.cuff.y.ok$test_stat>0 & tx.sex.cuff.y.ok$test_stat<=20))] = 1-cuff.p.tmp[which(tx.sex.cuff.y.ok$test_stat<(-20) | (tx.sex.cuff.y.ok$test_stat>0 & tx.sex.cuff.y.ok$test_stat<=20))]
# negative p-values
cuff.p.tmp[which(tx.sex.cuff.y.ok$test_stat>20 | (tx.sex.cuff.y.ok$test_stat<0 & tx.sex.cuff.y.ok$test_stat>(-20)))] = cuff.p.tmp[which(tx.sex.cuff.y.ok$test_stat>20 | (tx.sex.cuff.y.ok$test_stat<0 & tx.sex.cuff.y.ok$test_stat>(-20)))] - 1
head(data.frame(tx.sex.cuff.y.ok, cuff.p.tmp),100)
cuff.p.tmp.null = tx.null.cuff.y.ok$pvalue
# positives:
cuff.p.tmp.null[which(tx.null.cuff.y.ok$test_stat<(-20) | (tx.null.cuff.y.ok$test_stat>0 & tx.null.cuff.y.ok$test_stat<=20))] = 1-cuff.p.tmp.null[which(tx.null.cuff.y.ok$test_stat<(-20) | (tx.null.cuff.y.ok$test_stat>0 & tx.null.cuff.y.ok$test_stat<=20))]
# negatives:
cuff.p.tmp.null[which(tx.null.cuff.y.ok$test_stat>20 | (tx.null.cuff.y.ok$test_stat<0 & tx.null.cuff.y.ok$test_stat>(-20)))] = cuff.p.tmp.null[which(tx.null.cuff.y.ok$test_stat>20 | (tx.null.cuff.y.ok$test_stat<0 & tx.null.cuff.y.ok$test_stat>(-20)))]-1
head(data.frame(tx.null.cuff.y.ok, cuff.p.tmp.null), 100)
cuff.p.all = c(cuff.p.tmp, cuff.p.tmp.null)
cuff.exp = c(rep("sex",length(cuff.t)), rep("men",length(cuff.t.m)))
cuff.res = data.frame(t=cuff.t.all, exp = cuff.exp, p = cuff.p.all)
cuff.res = cuff.res[order(cuff.res$p,decreasing=T),]
cuff.p.rle = rle(cuff.res$p)
cuff.p.rle = data.frame(lengths=cuff.p.rle$lengths,values=cuff.p.rle$values)
for(k in which(cuff.p.rle$lengths>1)){
inds = which(cuff.res$t==cuff.p.rle$values[k])
cuff.res$exp[inds] = sample(cuff.res$exp[inds])
}
cuff.pct = NULL
for(i in 1:nrow(cuff.res)){
cuff.pct[i] = sum(cuff.res$exp[1:i]=="sex")/i
}
p.percentile = NULL
for(i in 1:length(cuff.res$p)){ p.percentile[i] = 1-i/length(cuff.res$p)}
edger.results$exp = rep("sex",nrow(edger.results))
edger.results.men$exp = rep("men",nrow(edger.results.men))
edger.fc = c(edger.results$logFC,edger.results.men$logFC)
esp = edger.results$PValue
esp[edger.results$logFC<0] = -esp[edger.results$logFC<0]
enp = edger.results.men$PValue
enp[edger.results.men$logFC<0] = -enp[edger.results.men$logFC<0] # has positive fold changes indicating overexpression in men
edger.p = c(esp, enp)
edger.exp = c(edger.results$exp, edger.results.men$exp)
edger.logCPM = c(edger.results$logCPM, edger.results.men$logCPM)
edger.res = data.frame(fc = edger.fc, exp = edger.exp, p=edger.p)
edger.res = edger.res[is.finite(edger.logCPM),]
edger.res = edger.res[order(1/edger.res$p,decreasing=T),]
# handle ties (randomly sample labels):
p.rle = rle(edger.res$p)
p.rle = data.frame(lengths=p.rle$lengths,values=p.rle$values)
for(k in which(p.rle$lengths>1)){
inds = which(edger.res$p==p.rle$values[k])
edger.res$exp[inds] = sample(edger.res$exp[inds])
}